Edge & Cloud are actually classic IT topics. Can you briefly outline the status quo of technologies in IT?

Thomas Frahler is Business Lead IoT at Microsoft Germany.

© Microsoft

Thomas Frahler: Let's start with the cloud: 84% of German companies already use cloud computing, according to the latest "Cloud Monitor 2022" study by KPMG and Bitkom Research. And although there have been very high growth rates in the past, the pace of growth is not slowing down. The spectrum of use within IT is now broad - it ranges from the outsourcing of individual infrastructure components, databases or services to the complete hosting of applications for internal or external use. No-code or low-code solutions contribute to this, as they allow IT or other specialist departments to quickly develop and provide their own applications hosted in the cloud.

With edge computing, the approach is reversed: data, applications, services and processes are not hosted and processed centrally as in the cloud, but in a decentralized manner. This makes it easier to meet requirements such as real-time data processing.

However, there is no rigid demarcation between the two models; they can now be easily combined. This means that cloud added values such as scalability can also be transferred to the edge. In concrete terms, this means, for example, that only selected components can be moved to the cloud and run there - i.e. applications, device and security management, data analysis including modeling or the training of AI algorithms. In the opposite direction, updates, for example, can be rolled out from the cloud to all relevant edge devices at scale.

The technologies are now also increasingly being discussed in OT, i.e. in the manufacturing environment. What is the difference between the two application areas of IT and OT?

Frahler: Firstly, in terms of organization, because IT is responsible for issues that affect the entire company, while OT is normally responsible for and assigned to a specialist department. Based on the ISA95 model, IT is located at the corporate level, i.e. level 5, where it is responsible for the operational and strategic management of the entire company. This includes core systems such as enterprise resource planning, but also areas such as communication and cyber security. It is also responsible for the operational management of the entire production process, including production planning, at production level, i.e. level 4.

At the levels below, OT is responsible for the operational control of production lines - level 3 - for example using SCADA solutions for the technical processes in automated production. In addition, there is machine and system control, i.e. level 2, and the product level - level 1. Gartner defines OT as hardware and software that detect or cause a change through the direct monitoring and/or control of physical devices, processes and events in the company - that describes it very well.

How well are the topics of Edge & Cloud already being accepted in OT? What acceptance problems are you struggling with?

Frahler : We have not identified any acceptance problems. As already mentioned, the results of the "Cloud Monitors 2022" study show a high level of adaptation and acceptance of cloud computing. However, despite the great interest on the part of OT, adaptation is not yet as advanced as in IT.

There are several reasons for this: In the past, cloud offerings have tended to be aimed at addressees in the IT sector. In addition, OT-specific requirements have only been specifically taken into account in recent years - either by translating and adapting corresponding offerings from the IT field for OT or by developing them directly for OT. Added to this is the complexity of connecting to company-wide systems and adapting processes, as well as the limited availability of tailored offerings.

And finally, the lack of skilled workers is another delaying factor, but it is an economic problem beyond OT.

What is your experience with your customers when introducing Edge & Cloud topics?

Frahler: As part of the digitalization and networking of processes and production workflows, overlaps between IT and OT cannot be avoided. This is because networked production systems or robots connected to the cloud require a cybersecurity strategy that is embedded in the company-wide security strategy. As a rule, this is the domain of IT. However, both areas must assert their requirements during implementation. We are therefore seeing that responsibility is increasingly being shared between IT and OT and that combined teams are being created accordingly.

Due to the complexity and many options associated with edge and cloud technologies, it is important for companies to plan for the long term when introducing them and to constantly reprioritize and adapt. In short, successful teams and projects pursue a long-term vision, but implement it in agile steps. This ensures that investments generate a return on investment more quickly.

To prevent the formation of silos, we recommend using open source and open standards, data models and platforms in the IIoT context. It also makes sense to identify implementation goals beyond your own specialist area. In this way, further added value and higher ROIs can be achieved through networked data. One example of this is the current win-win scenarios that arise when IIoT and sustainability are considered together. As evidence, the IDC study "Industrial IoT in Germany 2022", based on 250 companies surveyed, lists five top topics: CO2-neutral sites, re-manufacturing, material passport and lifecycle and product lifecycle management.

How can the problems of understanding/responsibilities between IT and OT be tackled?

Frahler : To ensure the implementation of IIoT application scenarios in the required quality, targeted communication and cooperation between IT and OT teams is essential. This is because IIoT starts at the intersection of both worlds, but also penetrates deep into both - after all, it extends from the sensor directly on or in the industrial asset to the central data centers and cloud environments of IT. This is why the integration of IT and OT, alongside a holistic data strategy, forms the foundation for IIoT.

For this integration, IDC recommends specific measures in the study that ensure an intensive transfer of knowledge. Initially, for example, the transfer of personnel to other departments could be considered. However, the aim should be to create combined teams or departments. These new digital engineering teams then act as a central IIoT instance and intermediary. However, clear guidelines and close support are required until they are in place: the biggest challenge in IT/OT integration is complexity.

What other challenges usually arise during implementation?

Frahler : In addition to the complexity of implementation, customers often encounter further difficulties. On the technical side, existing IT and OT structures prove to be obstacles, as do a lack of interoperability and homogeneity as well as scaling problems. On the personnel side, one of the most common difficulties is that the ability to use operational data sensibly has yet to be developed, as well as the re-skilling and up-skilling of employees that may be required.

How do you work out the question with your customers: Which data needs to be stored centrally (cloud) or decentrally (edge)?

Frahler: This is an iterative process that starts with the "why": First, it must be clear which goals are to be achieved and how these goals serve the corporate strategy and the achievement of overarching objectives.

The application scenarios are then defined in a brainstorming phase. They are evaluated in terms of feasibility and costs and then prioritized in a long-term implementation roadmap. The question of the overall architecture must also be clarified in this phase. It is often the case that a new data platform needs to be created first and governance needs to be reorganized. Among other things, this involves determining which data is stored in which scenario and how.

Cost-effectiveness, in particular productivity and efficiency, and of course security are usually also among the aspects examined. For example, caching an image data stream locally on the edge and analyzing and evaluating it there is cheaper, more efficient and more secure than mapping the entire process in a constant feed to the cloud. However, it makes sense to transfer the results of the analyses, including anomalies, critical incidents or other special cases, to the cloud at regular intervals. Models can then be retrained there and made available to all devices as an update.

Ultimately, the individual use case and the needs of the company are always decisive, which is why I avoid generalizations. However, your question can be answered in concrete terms: data for processing in real time, such as production-critical machine and system data, should ideally be as close to the device as possible in order to keep latencies to a minimum. And that means at the edge.

Applications in practice

Can you also quantify your project experience to date - back it up with concrete figures?

Frahler: Yes, there is already a wealth of well-documented applications in practice.

For example, the utility company E.ON has developed the sustainable district heating and cooling service E.ON ectogrid, which maximizes the reuse of energy within the system so that the amount of energy supplied can be reduced by up to 75 percent. This service uses the E.ON ectocloud digital platform for dynamic control and monitoring in buildings of different sizes. It records the measured values supplied by around 10,000 sensors - around 2.5 million every day - and also integrates weather forecast and energy market data. Based on this information, ectocloud uses machine learning algorithms to adjust heat usage at short intervals - the buildings are scored every 30 minutes. E.ON uses Azure Machine Learning, Azure Data Factory and Azure DevOps to link the processes so that the solution can be quickly rolled out internationally and flexibly developed further. This service shows what the combined use of detection, prediction and optimization algorithms can achieve in conjunction with control via IoT, while also illustrating the associated complexity and high demands on IT.

There is no shortage of other examples. For example, the global energy service provider Techem Energy Services uses daily data collection to obtain up-to-date information on water consumption, heat supply and usage by networking its more than 39 million radio recording devices in over 20 countries. Techem Energy Services uses this database to generate information with the help of the Microsoft Cloud, Azure Data Explorer and Azure IoT Hub in order to reduce energy consumption in a targeted manner. The company has calculated that this saves around 8.7 million tons of CO2 per year.

At fan manufacturer Ziehl-Abegg, an IoT platform developed with T-Systems MMS monitors its own and third-party devices that are used in important infrastructure such as operating theatres, production facilities and data centers. Packaging solutions provider Multivac, headquartered in Wolfertschwende, Germany, in turn reduces the changeover times of its latest generation of machines and simplifies maintenance thanks to smart services that process data in Microsoft Azure. In order to optimize the maintenance process of rotating machines in production, Bosch has developed an Integrated Asset Performance Management (IAPM) solution for the cloud based on Microsoft Azure with a digital twin. As part of predictive maintenance, it not only takes current sensor data into account, but also creates a holistic data situation from all relevant machine data so that the symptoms of impending wear and tear can be detected at a very early stage and the resulting failures can be effectively avoided. Bosch received the Microsoft Intelligent Manufacturing Award 2021 in the "Add Value!" category for this project. The "Oscar of the manufacturing industry" is awarded by Microsoft together with the management consultancy Roland Berger.

I could name a whole host of other cross-industry applications that demonstrate the benefits of IT/OT integration via the cloud. But it has long been recognized by the industry - the high adoption rate of cloud computing and the unabated growth show impressively that it is already tapping into the potential.